Experimental setup - Universiteit Utrecht

Chapter 3

Experimental setup

The 3 ~He(~e, e′n) experiment described in this thesis was performed using theaccelerator facilities of NIKHEF (Nationaal Instituut voor Kernfysica en Hoge-Energie Fysica), see figure 3.1. From the Polarized Electron Source (PES) pulsesof electrons were injected into the Medium Energy Accelerator (MEA). Afteracceleration to 720 MeV they were stacked in the Amsterdam Pulse Stretcher(AmPS) ring. In the target area (Internal Target Facility, ITF) the beam wassteered through an open ended storage cell in which polarized 3He gas wasinjected. Various detectors were placed around the target cell in order to de-tect, identify and analyze scattered electrons as well as hadrons emerging fromcollisions of the electrons in AmPS with the injected 3He atoms.In this chapter we describe the various elements of this setup in more detail.

3.1 The MEA/AmPS accelerator facility

The Medium Energy Accelerator was constructed in the seventies as a 500 MeVlinac for electrons. The duty factor was about 1%. In the late eighties the de-cision was taken to extend the accelerator with the Amsterdam Pulse Stretcherring (AmPS). In this ring a continuous current could be stacked by injectingwith MEA pulses of electrons with a length of one, two or three times the rev-olution time of the ring (0.7 µs).

In stretcher mode a high duty factor (> 80%) beam of up to 12 µA wasextracted from the ring and projected onto an external target in the Emin ex-perimental hall; the MEA repetition rate was typically 50-150 Hz. With the

39

40 CHAPTER 3. EXPERIMENTAL SETUP

AmPS

MEA

CBP

Siberian Snake

ESC

thermionic gun

Emin

ITF

PES

50 m

Figure 3.1: Schematic map of the Medium Energy Accelerator and the Amster-dam Pulse Stretcher ring.

100

time in seconds

curr

ent i

n A

mPS

200 300 400 500 600 700 800 900 1000

0.15

0.1

0.05

taking dataHV

dow

n

inje

ctio

n

inje

ctio

nH

V u

pbeam

kill

HV

up

taking data HV

dow

nbe

am k

illin

ject

ion

HV

up

taking data

Figure 3.2: Current in AmPS versus time, in storage mode. Some aspects ofexperimental procedures are added to the graph. One “fill” (removing the oldbeam, stacking current, ramping up the detectors, taking data and rampingdown) takes typically 5-7 minutes. Between fills the target polarization wasflipped. After each run (about 10 fills) the electron helicity was flipped. Sincethe radiation produced during injection might ruin the sensitive componentsof some of the detectors, their high voltage (HV) had to be ramped down (up)before (after) each injection phase. Discarding (“killing”) the old beam betweeninjections is necessary for polarized electron beams since the polarization hasa finite lifetime. It is particularly important between different runs, when theelectron helicity is flipped.

3.1. THE MEA/AMPS ACCELERATOR FACILITY 41

high luminosity, which was of the order of 1036cm−2s−1 (= pb−1s−1), processeswith small cross sections like two nucleon knockout reactions could be studied(see, e.g. [83]).

In storage mode several pulses were injected (at 1 Hz) into the AmPS ring.After each injection the beam rapidly damped to an equilibrium trajectory. Theamount of stacked current in the ring had roughly the same limitations as instretcher mode. But in contrast to stretcher mode, where every electron passedonly once through the (external) target, the beam was ‘recycled’ with a rateof 1.4 MHz so that although the injection rate was two orders of magnitudelower than in stretcher mode, a four orders of magnitude greater current (upto 250 mA) illuminated the target.

In order to preserve the beam quality, the thickness of the internal targethad to be extremely small, less than about 1015 atoms per cm2. At higher den-sities the beam blows up to the extent that it is impossible to refocus it andmost of the electrons will scatter from the beam pipe and will be lost. Possi-ble internal targets are fibers, pellets [61] and low-pressure gas. In the internaltarget in the AmPS a carbon fiber target was used for the calibration of theelectron spectrometer. For physics experiments gas targets were employed.The gas flow into the scattering chamber was limited to 1018 atoms/s by thevacuum requirements of the AmPS ring (10−8 mbar) and the speed of the turbopumps. In order to maximize the luminosity within these limitations, the gaswas guided through a cooled tube of 15 or 20 mm in diameter around the beam,which yielded target densities of the order of 1014 atoms per cm2.

Thus the experiments in the internal target facility (ITF) could be performedwith luminosities of the order of 1032cm−2s−1 (= 0.1 nb−1s−1). The very low tar-get density enabled the detection of light recoiling nuclei and the backgroundrates were comparatively low.

In order to maintain the polarization of a beam stored in AmPS a set ofsuperconducting solenoids was installed opposite to the location of the inter-nal target. These solenoids compensated the spin precession that occurred inthe four bends. The precession angle α after one revolution depends only onthe electron energy: α = 2πE/Emagic , where Emagic = 440.65 MeV. In princi-ple these solenoids (the ”Siberian Snake”) enabled a polarized electron beamin both storage mode and stretcher mode. However, the polarized electronsource was not designed to run at higher rates than 1 Hz and hence could notdeliver the amount of current needed for running in stretcher mode. More-over, in order to get a proper electron spin orientation at the location of theexternal target, solenoids would be necessary in the extraction line [88]; they


were however never installed.Figure 3.2 displays the typical time structure of beam current and data tak-

ing. The beam current decayed (by interactions with the target, rest gases andthe beam pipe and due to synchrotron radiation) with lifetimes of the order of300 seconds (with target gas) and 1200 seconds (without target gas in the ring).

3.2 Polarized electrons

3.2.1 Polarized electron source

Due to synchrotron radiation a stored electron beam will (in principle) be-come polarized perpendicularly to the plane of the ring. This is the so-calledSokolov-Ternov effect [14]. The equilibrium polarization of 92% is reached af-ter 15.8× ρ2LE−5 seconds, where ρ is the bending radius (in meters) of thebeam path in the dipole magnets, L is the circumference of the ring (in meters)and E is the electron energy (in GeV). For AmPS full self-polarization wouldtake more than a day whereas the typical lifetime of the beam with no gas inthe internal target is about 20 minutes. Hence, for nuclear physics experimentsthis polarization mechanism is not useful. However, after stacking of a max-imum current of some 250 mA it takes several hours until it has decayed toless than 10 mA. By that time there should have been a significant polarizationbuild-up already. Measurements of this longitudinal self-polarization processhave been performed with AmPS and will soon be published [101].

In the absence of a useful self-polarization mechanism a Polarized ElectronSource (PES) was developed and constructed at NIKHEF [71]. With circularlypolarized laser light electrons of one particular helicity are photo-emitted froma strained layer crystal and steered through a so-called Z-manipulator, whichconsists of a series of solenoids and electrostatic bends. By tuning the solenoidsthe spin could be oriented in any direction. During experiments the orientationwas tuned such that in AmPS at the location of the internal target the spin wasprecisely (anti)parallel to the beam direction.

3.2.2 Polarimetry

The degree of polarization directly after the Z-manipulator was measured witha Mott polarimeter, for which the electron spin had to be oriented vertically. Foreach crystal this was done directly after installation, before removal and a few

3.2. POLARIZED ELECTRONS 43

day in the year 1998

PM

ott [

%]

I II III IV V40

50

60

70

80

40 60 80 100 120 140 160 180 200

Figure 3.3: Electron polarization measured with the Mott polarimeter. The hor-izontal bars indicate the various stages in the experiment: I Commissioning ofthe polarized electron beam and measurement elastic asymmetry; II Tuningand calibration of the Neutron Detector and the range telescope; III Measure-ment of A′x; IV Measurement of A′z; V Measurement of A′x.

times in between, typically once every few days. The polarization thus foundis plotted versus time in figure 3.3.

The electron polarization was also measured in AmPS with a ComptonBackscattering Polarimeter (CBP) [68, 89], see figure 3.4. The stored electronbeam was irradiated with circularly polarized laser light in the beginning ofthe first bend behind the internal target. After Compton scattering the photonshad a momentum vector in a narrow cone around the tangent to the electronbeam. They were counted in a CsI crystal which was surrounded by a 10 cmthick lead cylinder and preceded by a concrete collimator and a sweeping mag-net. By comparing the Eγ spectrum for right oriented with that for left orientedlaser light the degree of polarization of the electron beam could be determined.

Ideally, the electron polarization would be monitored continuously, but be-cause of bremsstrahlung and synchrotron radiation backgrounds the CBP mea-surements could only be performed at small beam currents (< 20 mA) withoutgas in the target cell. Hence these measurements had to be performed sepa-rately. For the determination of the asymmetries in the data analysis the polar-ization as measured with the Mott polarimeter has been used.

Theoretically, the degree of polarization should not depend on the amountof beam current. However, in our experiments we found strong indicationsthat there is such a dependence. This is discussed in chapter 5.


laser

c� hopperlenses

Pockels cellλ

�/4-plate

mirror

2 mirrors

positioning s� ystem

a� nalyzing s� ystem

γ� -detector

magnet

internal t

�arget laser light

Compton photons

e� lectrons

interaction region

λ�

/2-plate

BC

D

A

Figure 3.4: The Compton Backscattering Polarimeter in AmPS. The laser lightwas polarized by means of a quarter-wave plate and a Pockets cell (A). A sys-tem of mirrors, all under 45◦ with respect to the laser beam, guided the beaminto the first section of the bend directly after the internal target (C). The inter-section of the laser beam with the electron beam was optimized with the beampositioning system (B). The photons that do not undergo Compton scatteringwere reflected into the analyzing system (D). This system consisted of a powermeter preceded by a linear polarizer and a rotating half-wave plate; the ratioof the oscillation amplitude and the offset of the power is equal to the degreeof circular polarization of the laser light.

3.3. POLARIZED 3HE TARGET 45

3.3 Polarized 3He target

A source of polarized 3He atoms was developed and constructed at the VrijeUniversiteit [79]. Figure 3.5 shows part of the target setup. In a glass pump-ing cell mounted directly above the target cell a fraction of the atoms of 3Hegas at low pressure (1 mbar) was excited by an RF discharge to the metastable23S1 state. The magnetic holding field, generated by a set of Helmholtz coils(not shown in the figure) around the scattering chamber, defined the quantiza-tion axis. By irradiating the cell with left (right) circularly polarized laser lightparallel to the magnetic field, the atoms with total magnetic quantum numbermF > 0 (mF < 0) were excited to 23P0,1,2 states. The atoms decayed back to the23S1 with either positive or negative mF. After some time only the states withmF < 0 (mF > 0) were populated. The polarization was transferred from themetastable 23S1 atoms to the ground state (11S0) atoms by so-called metasta-bility exchange collisions; and since in the ground state of a 3He atom the totalangular momentum of the two S-state electrons is zero, the polarization wasin that state completely carried by the nucleus. The resulting polarization wasmeasured by monitoring the polarization of the fluorescence light emitted bythe 31D2→ 21P1 transition in the RF discharge.

Via a feed tube the 45-50% polarized gas flowed with a rate of 1017 atomsper second into the target cell, resulting in a target thickness of 0.7× 1015 atomscm−2. In order to achieve this density the target cell was cooled to 17 K.

The direction of the polarization (ϑ∗, ϕ∗) was defined by the magnetic fieldgenerated by the three pairs of Helmholtz coils and, correspondingly, the di-rection of the laser light. There were three different settings:

• Elastic scattering: ϑ∗s = 98◦, ϕ∗s = 0 (ϑ∗lab = 27◦, ϕ∗lab = 0), close to thedirection of the BigBite spectrometer, which was located at ϑBB

lab = 30◦

(ϑlabq = 71◦). This reaction channel served to check the luminosity and

the product of beam and target polarization. The chosen direction of thetarget polarization maximizes the experimental asymmetry.

• Parallel (measurement of A′z): ϑ∗s = 0◦ (ϑ∗lab = 56◦, ϕ∗lab = 180◦) with theBigBite spectrometer positioned at an angle of ϑBB

lab = 40◦, so in this mea-surement the target polarization is parallel to the central value of the mo-mentum transfer q. However, when a field map was made after the datafor this set were taken, it turned out that due to a wrong polarity of theDC current power supply of one of the Helmholtz coil pairs the actualdirection was ϑ∗lab = 28◦, ϕ∗lab = 180◦ (ϑ∗s = 28◦, ϕ∗s = 0◦). So instead of


pumpingCell

copperrod

storage cell

wakefieldsuppressor

coldhead

3He

Figure 3.5: Some components of the target setup. By cooling to 17 K the targetdensity increased with a factor of four. Copper braids conduct the heat fromthe target cell to a 40 × 40 mm2 copper bar, connected to a 30 Watt coldhead.The connection is curved in order to allow for the Helmholtz coils (not shown).

3.4. DETECTORS 47

A′z a linear combination A′z cos(28◦)− A′x sin(28◦) was measured. In the3He(e, e′n) channel A′x is an order of magnitude less than A′z so that inthe neutron channel the effect of this mistake was a reduction of the mea-sured asymmetry by a factor of cos(28◦) = 0.88.

• Perpendicular (measurement of A′x): ϑ∗s = 90◦, ϕ∗s = 0◦ (ϑ∗lab = 34◦, ϕ∗lab =0◦ and still ϑBB

lab = 40◦).

Data with ϑ∗s = 90◦, ϕ∗s = 90◦ (ϑ∗lab = 90◦, ϕ∗lab = 90◦), comprising a measure-ment of A0

y, were taken in 1997. The analysis of this experiment may be foundin [79].

3.4 Detectors

As mentioned above, the 3 ~He(~e, e′X) experiment was performed in three stages:elastic scattering, and measurements of A′z and A′x. For the measurement ofelastic 3He(e, e′3He) scattering the Recoil detector had to be on the oppositeside of BigBite (figure 3.6(a)). For the A′z and A′x measurements the Recoil De-tector was mounted on the same side as BigBite in order to measure protonsand deuterons optimally with the range telescope, (figure 3.6(b)).

3.4.1 The BigBite electron spectrometer

The BigBite electron spectrometer consists of a dipole magnet followed bytracking and particle identification detectors. The whole system is mountedon a platform which can be rotated to angles between 25◦ and 90◦ (in steps of5◦) around a pivotal point fixed below the target center via a system of air-pads.In figure 3.7 all basic components are shown in a side view.

A particle emanating from the target enters the spectrometer at 99 cm fromthe target center through the 25 cm wide mouth of a wedge-shaped dipolemagnet1. The nominal field intensity of 0.92 Tesla deflects a 500 MeV electronentering along the optical axis by 25◦.

After deflection in the magnet the electron traverses a pair of multi-wiredrift chambers. The chambers are 70 cm apart and have active areas of 140×35 cm2 and 200× 50 cm2, respectively.

In each wire chamber, the dispersive direction of the track is measured intwo wire planes, with a pitch of 20 mm, shifted half a pitch with respect to each

1Constructed in the Budker Institute for Nuclear Physics, Novosibirsk, Russia


1 m

70 o

e

recoil detector

30 o

BigBite

(a) Elastic 3 ~He(~e, e′3He)

40 o

56

115

1 m

e

Neutron Detector

recoil detector

o

o

BigBite

RangeTelescope

(b) A′z and A′x for 3 ~He(~e, e′X)

Figure 3.6: Schematic maps of the detector configuration (only roughly onscale).

other. The time difference between the trigger of the scintillator (see below)and the anode wire hit scales with the drift time of the ionization charge tothe wire. This drift time, ranging from zero to several hundred ns, constrainsthe dispersive coordinate to within 180µm, apart from a reflective ambiguitywhich is removed when at least three (out of four) wire planes fired.

The nondispersive coordinate is measured by copper strips on the outer(cathode) planes of each drift chamber. The strips are 4.08 mm wide and arelaid down on mylar foil with a pitch of 5.08 mm. Comparison of the amount ofcharge collected by a strip with the charge collected by the neighboring stripsconstrains the nondispersive coordinate of the track to within 100µm.

The coordinates of the hits in the drift chambers together with the assump-tion that the particle emanated from the beam line fix the track of the particlein the field of the dipole. The bending radius of this track and the field strengthuniquely determine the momentum.

The timing for the drift chambers is set by a 200× 50× 1 cm3 scintillatordirectly after the second chamber. On each side there are two photomultipliers,of which at least one must fire, in order to produce – with a mean timer –a trigger that is (almost) independent of the position in the scintillator. This

3.4. DETECTORS 49

scintillator

air pads

drift chambers

fieldclamp

magnet

Cerenkovˇ

Figure 3.7: Components of the Bigbite electron spectrometer.


Quantity Acceptance ResolutionMomentum 250 . . . 900 MeV/c 0.84%

θ ±80 mrad 3-5 mradφ ±300 mrad 3-5 mrad

vertex ±300 mm 3.2 mmtiming 1 ns

Table 3.1: Acceptance and resolutions (standard deviations) of the BigBite spec-trometer.

trigger also determines the timing of the BigBite arm trigger that is sent to theCoincidence Detector (see figure 3.15).

The top component of the electron spectrometer is a Cerenkov detector con-sisting of a 210× 50× 24 cm3 stack of aerogel blocks, read out via a diffuselyreflective light box by twelve 5 inch photomultipliers [67]. The refractive in-dex of aerogel is 1.05 and the material has an small density so that the en-ergy loss by atomic collisions is negligible. Charged particles with a speedgreater than c/1.05 = 0.95c (where c denotes the speed of light in vacuum)produce Cerenkov light. Pions with such a speed have a momentum greaterthan 435 MeV/c. The cross section to produce such fast pions with 720 MeVelectrons is negligibly small.

In the environment of the internal target hall only electrons and cosmicmuons could produce a detectable signal in the Cerenkov detector. The mo-mentum distribution of the cosmic radiation has its maximum around the ze-nith. The assumption that these tracks would originate from the beam linetherefore results for most of the cosmics in a small bending radius in the dipoleand hence a very low momentum (peaking at the minimum of the acceptance,around 250 MeV/c.

Table 3.1 summarizes the acceptance and resolutions of the BigBite spec-trometer. For more extensive descriptions of its design, calibration and prop-erties the interested reader is referred to [70, 86, 66, 67].

3.4.2 Recoil detector

In contrast to fixed (liquid, solid) targets, an internal gas target is so thin thatlow-energy recoiling nuclei can escape from the target and be detected. Thisenables the measurement of reaction channels that would otherwise be either

3.4. DETECTORS 51

impossible or extremely difficult. To exploit this advantage of the internal tar-get, the Recoil Detector was designed and constructed by the Vrije Universiteitgroup within NIKHEF.

The Recoil Detector was contained in a low-pressure (7 mbar) 300× 308×150 mm3 box mounted directly on the scattering chamber, either on the oppo-site side to the BigBite at 70◦ or on the same side at 115◦. A 99% transparentgrid before the entrance of the detector suppressed the electro-magnetic pickupinduced by the RF field of the beam.

A recoiling particle from the target (e.g. a deuteron) with sufficient energytraversed a 0.9µm mylar foil that separated the detector atmosphere from theAmPS vacuum, a low-pressure wire chamber, two silicon strip detectors (SSD)and a scintillator (see figure 3.8). However, because of a high rate of low-energyelectrons from Møller scattering, the wire chamber could only function prop-erly when permanent magnets were mounted on top and below the entranceof the detector to deflect those electrons to smaller angles. Since the gradientsof the fringe fields of these magnets would have ruined the target polarizationthe wire chamber of the Recoil Detector was not used.

Each of the two SSDs had three 50× 50 mm2 segments, divided in 16 strips.On the first (second) layer the 0.1 mm (0.5 mm) thick strips ran vertical (hor-izontal), giving horizontal (vertical) position information. The distance of thefirst silicon layer to the target center was 21.4 cm so that with a vertex resolu-tion of 0.5 cm an angular resolution of 2◦ could be obtained. The light nuclei(1,2,3H, 3,4He) stopping in the second layer could be cleanly identified with the∆E/E method. The maximum detectable kinetic energy of a 3He nucleus in thiscase was 31 MeV, or 83 MeV if it stopped in the 60× 180× 5 mm3 scintillator.

For more extensive descriptions and analyses the reader is referred to [78,63, 70, 75, 103, 87].

3.4.3 Range telescope

Figure 3.9 shows the layout of the range telescope, which was positioned at69.0 cm from the target center at an angle of 56◦. It consists of two wire cham-bers and an array of two thin layers of plastic scintillator (30× 50× 0.2 cm3)and fourteen thicker layers (30× 50× 1.0 cm3).

The multi-wire proportional chambers have three wire planes – in the X, Yand Θ (= 45◦) direction – with a pitch of 6 mm. A 1 mm aluminum plate wasplaced on the scattering chamber over the exit foil to protect the wire chambersagainst low energy background.


PMT

Scintillator

Lightguide

Wire Chamber

SiY

SiX

Charge sensitive preamplifiers

Entrance Foil

100 mm

100

mm

Figure 3.8: Components of the Recoil Detector.

18.5 cm

Plastic Scintillators

Wire Chambers

Phototubes

50 cm

Figure 3.9: Components of the range telescope.

3.4. DETECTORS 53

The trigger was defined by the coincidence of hits in the second and thethird scintillator layer; in this way the background was reduced but the deu-terons of interest (kinetic energy greater than 45 MeV) could still be detected.Deuterons with more than 200 MeV kinetic energy punched through the lastscintillator layer. The kinetic energy range for protons was 33-150 MeV. Theangular acceptance was 56◦ ± 9◦ in the polar angle and ±15◦ in the azimuthalangle.

This detector was used in several earlier experiments at NIKHEF [60, 65]and in Saskatoon [55]. The full analysis of the 3 ~He(~e, e′p) and 3 ~He(~e, e′d) reac-tion channels may be found in [85].

3.4.4 Neutron detector

Neutron detection and Time-of-Flight

In order to detect a particle it has to interact with the detector material. Acharged particle mostly interacts electromagnetically. If it has sufficient energyit leaves an ionization trace in any medium except vacuum. There are severaltechniques to amplify and detect ionization charges. In suitably chosen mediathe position of these charges can be reconstructed with resolutions better thana µm, whereas the amount of charge in the trace may be used in the determi-nation of the energy and the identity of the particle.

Neutrons do not have a net electric charge. They do have nontrivial electro-magnetic structure, but the electromagnetic form factors are not large enoughfor a neutron to cause an ionization trace.

Only the properties of the strong interaction are relevant for the design ofa detector for medium energy neutrons. But although αs is (at nuclear scales)two orders of magnitude greater than αe, the density of scatterers (nuclei in-stead of electrons) is less and the strong interaction has a comparatively shortrange. The interaction probability for a 100 MeV neutron in a medium with adensity of 1 g/cm3 is of the order of 1% per cm.

In the collision with a nucleus a random fraction of the kinetic energy ofthe neutron is transferred to (fragments of) the nucleus with which it inter-acted (see the subsection on efficiency, page 57). Most of these fragments arecharged and hence ionize the detector material (provided their kinetic energyis sufficient).

Therefore, a measurement of the total energy deposited in principle only


yields a lower limit for the kinetic energy of the neutron2. If the detector isso large that the neutron and the secondary scattering fragments are, with ahigh probability, completely stopped, then the deposited and the initial kineticenergy will be strongly correlated. However, for an experiment at intermediateenergies with a plastic scintillator detector this would require several meters ofthickness. Such a detector is expensive, impractical and would suffer a lot frombackground radiation.

Instead of by measuring the deposited energy, one may determine in a co-incidence experiment the kinetic energy of a nonrelativistic neutron (or anyother particle) in an alternative way, namely with the Time-of-Flight technique(TOF). If of one of the other products of the primary scattering (in our case thescattered electron) the momentum and the trigger time are accurately known,then the vertex and the instance of the primary interaction can be reconstructed.This information together with the impact position and the trigger time inthe TOF detector, determines the distance-of-flight and the time-of-flight andhence the speed of the particle (see figure 3.10).

In order to use a TOF detector for neutron detection it is usually precededby one or more thin charged particle detectors (”veto layers”); when the TOFdetector fires but the veto layers do not, one assumes a neutral particle wasdetected. This may be either a photon or a neutron. Neutral mesons (usually)do not live long enough to travel the typical distances (several meters) to TOFdetectors3. A good timing resolution is required to separate the photons (β =1) from the fastest possible neutrons in the experiment (β < 0.8 in our case).Particles that do not originate from the target location may enter the detectorwithout passing the veto layers. This low energy background radiation maybe eliminated by requiring a minimum energy deposit, at the expense of somedetection efficiency.

For various reasons scintillator plastic is the natural choice for the detectormaterial, since it has a reasonable density (and hence reasonable interactionprobability for neutrons), is transparent and relatively cheap (which allowslarge detector volumes) and has good timing properties (the primary prerequi-site for a TOF detector). In the following a ”neutron detector” is understood tobe a TOF detector constructed of scintillator blocks equipped with photomul-tiplier tubes, preceded by thin scintillator layers that may serve as veto layers.

2Except when the scattering was elastic and the scattering angle is also known; then the mo-mentum of the neutron can be exactly reconstructed. This is the principle of the High AcceptanceRecoil Polarimeter detector (see refs. [51, 62]).

3The lightest neutral meson which does live long enough is the KL meson; which due to its massof 598 MeV does not play a role in the present work.

3.4. DETECTORS 55

1

2

0

(x , t

x , t( p , E( x , t) )

)

1

1

2

0

2

1

10

Figure 3.10: In order to measure the velocity of the particle detected at posi-tion x2 at time t2 the position x0 and time t0 of the primary vertex must becalculated from the coordinates and momenta of the particle in detector 2.

On the one hand one would like to construct a neutron detector as thickas possible in order to have a high detection efficiency, given the low interac-tion probability. On the other hand, the thicker the block the bigger the rela-tive uncertainty in the distance-of-flight and hence the poorer the momentumresolution. The relative uncertainty in the distance-of-flight (and that of thetime-of-flight as well) may be reduced by increasing the distance to the target,which, however, decreases the solid angle. Another way to improve the dis-tance resolution is to split the blocks in successive layers (which implies theinvestment in more complicated electronics and trigger logic).

The TOF detector in ITH

Figure 3.11 shows the geometry of the neutron detector without support struc-tures. The detector consists of two identical walls, each consisting of four tele-scopes, which consist of three scintillator bars. They were designed and con-structed by the Universiteit Utrecht as a component of the HARP detector [51],for which they were optimized to detect (recoil) protons.

The bars are 160 cm long, 20 cm wide and 20 cm (E), 1 cm (M) and 0.3 cm


Top view

pm tubes

lightguides

scintillators90 cm

Front view Side view20 cm

20, 21, 21.3 cm

160 cm

Figure 3.11: Schematic outline of the HARP scintillator bars used as TOF de-tector in ITH.

(D) thick, respectively. The scintillator material is BC4004 (Bicron Corp.). Thebars are wrapped in two layers of aluminized mylar (to improve the light col-lection efficiency) and 0.2 mm thick black PVC (to exclude any light from out-side the bar). The E bars are read out on both sides by 5 inch photomulti-plier tubes (Burle 8854). On one side they are connected by a simple (cube andcylinder) light guide while at the other end the light is reflected over 180◦ bya double-prism light guide; this was necessary in the design of the HARP de-tector. The veto layers (M and D) are read out by 2 inch photomultiplier tubes(Burle 8575), connected to the bars by adiabatic light guides, which on one sidemake a 180◦ bend just like for the E bar.

Four telescopes are mounted with plastic straps on a steel support structure,which can move on a rail system inside a support frame. The two scintillatorwalls in their respective support frames were put together on a steel frame withwheels, such that the center of the array was at beam height and the scintilla-tors faced in the same direction. The detector was positioned in the internaltarget hall as depicted in figure 3.6.

4Equivalent to NE-102A.

3.4. DETECTORS 57

Kinetic Energy [MeV]

Lig

ht [M

eVee

]

0

5

10

15

20

25

30

35

40

45

50

0 5 10 15 20 25 30 35 40 45 50

Figure 3.12: Amount of light, expressed in electron equivalent MeV, versus ki-netic energy (MeV), for protons (solid), α particles (dash-dotted) and electrons(dashed) stopping in NE-102 scintillating material.

The central angles of the detectors were chosen to match the central anglesfor quasi-elastic nucleon knockout from a 3He nucleus at Q2 = 0.2 GeV2/c2

with an incident electron energy of 720 MeV. At this angle, it was positionedas far away from the target as possible. The front face of the first wall was at2.1 m from the target.

Efficiency

For a good comparison of a theory to measured data, the predictions of thetheory must be ’folded’ over the acceptance via a Monte Carlo simulation, asdescribed in section 2.4. For asymmetry measurements, the efficiency of thedetectors does not affect the result directly, in contrast to the case of absolutecross section measurements. However, the efficiencies are weighting factors inthe Monte Carlo simulation and hence their variation (rather than their abso-lute values) still needs to be known accurately.

The efficiency of a neutron detector may be determined experimentally byirradiation with a well-defined neutron flux, e.g. with the neutron beam at thePaul Scherrer Institut (PSI) in Villigen (Switzerland) [35] or with the 1H(γ, π+n)


reaction, see [56, 57]. In the latter the measurement of the π+ momentum vectorfixes both the photon energy and the neutron momentum.

Alternatively the efficiency may be estimated by means of a Monte Carlosimulation using cross section tables from neutron-nucleus and np scatteringdata. The KSUVAX program [20] simulates neutron interactions in detectorsthrough “tracking” and “scattering” with either protons or carbon nuclei with aprobability determined from the cross section tables. The program was checkedfor neutron kinetic energies up to 300 MeV, several scintillator types and detec-tion thresholds and found to be reliable to within 10%. The following processesare taken into account:

np → np (elastic scattering)nC → nC (elastic scattering)nC → n′γCnC → α9Be (assumed isotropic)nC → nαααnC → np11BnC → nn11C

Knocked-out α particles and protons, as well as the scattered neutron, are sub-sequently tracked, until they either leave the detector volume or deposit (al-most) all their kinetic energy. The efficiency of the conversion of the depositedenergy to visible light is retrieved from a fit to data from light response mea-surements. The amount of light is expressed in the unit MeVee, i.e. the amountof light generated when a 1 MeV electron stops in the detector material5. Thefitting function has the form

L [ MeVee] = a1T(p,α) − a2

[1.0− exp(−a3Ta4

(p,α))]

(3.1)

where the coefficients ai depend on the particle (p, α) and on the type of scin-tillator material, while Tp is the kinetic energy (MeV) of the particle. So bydefinition for an electron one has a1 = 1 and a2 = 0. In figure 3.12 the graphs ofthe fit for protons and α particles are shown.

The maximum possible neutron detection efficiency for a given detectorgeometry is the ratio of the number of neutrons that interact with the activedetector material over the total number of neutrons that enter the detector. Theinteraction probability depends only on the detector material, the length of the

5For electrons with a kinetic energy greater than 100 keV the amount of light (the sum of theenergy of the photons) is a linear function of the energy deposited by the electron.

3.4. DETECTORS 59

0

5

10

15

20

25

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1Horizontal position [m]

Eff

icie

ncy

[%]

0

5

10

15

20

25

0 25 50 75 100 125 150 175 200Tn [MeV]

Eff

icie

ncy

[%]

Figure 3.13: Panel (a): Neutron detection efficiency versus position (thresholdat 12 MeVee). Panel (b): Neutron detection efficiency as a function of the neu-tron kinetic energy for several threshold values of the total amount of light:6 MeVee (short dashes), 9 MeVee (dot-dashes), 12 MeVee (thick solid line), 16MeVee (dots) and 20 MeVee (long dashes). The attenuation length of the lightin the bar travelling towards the phototubes is 3.2 m. The photomultiplierthreshold is 5 MeVee.

neutron track inside the detector volume and the kinetic energy of the neu-tron. Since the probability distribution fraction of the neutron kinetic energytransferred to the charged scattering fragments has its maximum at zero anddecreases steeply, the actual detection efficiency depends strongly on the min-imum amount of light that gives a good signal in the photomultipliers.

In the following, a “good signal” is a signal that exceeds the hardwarethreshold and survives the software cuts. The thresholds of the photomulti-pliers have been tuned as low as possible but such that the rate of triggersfrom electronic noise was still negligibly low. As remarked on page 54, lowenergy random background may be eliminated in the analysis by requiring aminimum amount of scintillator light produced.

In the KSUVAX simulation 104 points were generated in a 4-dimensionalphase space spanned by the z coordinate of the primary vertex position (trian-gular distribution), the hit position in the detector (flat distribution for the ver-tical coordinate, Gaussian-like distribution for the horizontal coordinate with


0

20

40

60

0 5 10 15 20 25

ND threshold [MeVee]

Eff

icie

ncy[

%]

0

10

20

30

0 5 10 15 20 25

ND threshold [MeVee]

Eff

icie

ncy[

%]

Figure 3.14: Efficiency without (a) and with (b) light attenuation and hardwarethreshold of 5 MeVee. Solid: 0 < Tn < 35; dashed: 35 < Tn < 50; dot-dashed:50 < Tn < 75; dotted: 75 < Tn < 125; very dotted: 125 < Tn < 200

.

a maximum in the center) and the neutron kinetic energy (Gaussian-like). Foreach point 105 neutrons were tracked. The light production was attenuated bya factor exp[−d/λ] where λ= 3.2 m is the attenuation length and d the distanceto the farthest photomultiplier; if the amount of light after attenuation did notexceed the photomultiplier threshold of 5 MeVee then the light production wasset to zero. The statistics of the light production for the complete sample of105 neutrons then provided the efficiency for any detection threshold in thatparticular point in phase space.

Figure 3.13 displays the dependence of the efficiency of the first wall of theneutron detector on the hit position and on neutron kinetic energy for severalthresholds. The slight decrease of the efficiency for more downstream positions(larger values for the horizontal position) is due to the extendedness of thetarget cell, which makes that for the upstream part of the detector the averagelength of the track of the neutron through the detector material is greater thanthat for the downstream part. Figure 3.14 displays the efficiency dependenceon the threshold for several Tn bins.

The detection threshold for the time-of-flight detector in the analysis of the3 ~He(~e, e′n) experiment was put at 12 MeVee (see chapter 4). As may be con-cluded from figure 3.13, the average detection efficiency for one wall is (with

3.4. DETECTORS 61

this threshold) equal to 18±1% for 50< Tn < 200 MeV. So within the context ofthe analysis of this experiment neutron detection efficiency may be taken to beconstant. For neutrons emerging from the target into the acceptance of the sec-ond wall the detection efficiency is equal to 18% + (100%− 18%)× 18% = 33%.

3.4.5 Electronics and data acquisition

The electronics of the BigBite spectrometer, the Recoil detector and the rangetelescope are described in [66, 78, 85], respectively. Figure 3.15 displays theprinciples of the trigger, readout and data acquisition for the neutron detector.The analog signals of the photomultipliers were carried by 40 m long co-axialcables from the detector to Hadron Digitizer Modules (HDMs) [46] outside theexperimental hall.

In an HDM the analog signal is split in three. One of the copies is sent over adelay line to a charge integrator (QDC), the other two are led into a low thresh-old differential discriminator and a high threshold leading edge discriminator,respectively. If both discriminator thresholds are exceeded, a zero-level triggeris released, where the timing is determined by the low threshold in order toreduce the walk effect. This trigger defines the start time for both the TDC andthe integration interval of the QDC of this channel.

In the data acquisition setup for the ITH neutron detector the triggers of theE bar channels were passed to a trigger 1A module, which contains a triggerlogic applicable for the Hadron detectors. By not using all input channels thistrigger logic simplified to an effective trigger logic as displayed in figure 3.15,so it released a first level trigger when the signals of two photomultipliers ofthe same E bar had both generated a zeroth order trigger.

A first level trigger defines the stop time for the TDCs in all HDMs and issent as an arm trigger (ATR) to the Coincidence Detector (CD). The CD collectsATRs from all four detectors and determines whether the ATRs are single hitsor part of a double, triple or quadruple coincidence.

Under generic experimental conditions the rates of some types of coinci-dence, in particular those of the single hits, are too high to be able to store allrelevant detector data without introducing an unacceptable amount of deadtime. Therefore, for each kind of coincidence (including single hits) a prescalerPcoinc may be defined so that the CD generates an Event Trigger (ETR) for onlyone out of every Pcoinc instances of that kind of coincidences. The ETR signal issent to all arms in the coincidence, which in return send all their informationfor that particular event to the Event Builder (EB) after which the data can bestored on disk and tape.


Storage foronline andoffline analysis

DART

NDEventTrigger(ETR)

BigBitepretriggergate

EventBuilder

RT

dat

a

Rec

oil d

ata

Big

Bite

dat

a

NeutronDetector

Arm Trigger(ND ATR)

DetectorCoincidence

BB

AT

R

RT

AT

R

Rec

oil A

TR

Recoil E

TR

RT

ET

R

BigB

ite ET

R

*

*

*

*

*

*

Trigger 1A

Trigger 1B

Hadron Digitizer Modules

TDC

QDC

TDC

QDC

TDC

QDC

TDC

QDC

TDC

QDC

TDC

QDC

QDC start

TDC start

TDC start

TDC start

TDC start

TDC start

TDC start

analog signal

analog signal

analog signal

analog signal

analog signal

QDC start

QDC start

QDC start

QDC start

QDC start

analog signal

TDC stop

TDC stop

TDC stop

TDC stop

TDC stop

TDC stop

4DF

4DR

4MF

4MR

4EF

4ER

4ER

4EF

3EF

3ER

14EF

14ER

5EF

5ER

Figure 3.15: Intuitive representation of ND electronics and data acquisition.The discriminators indicated with an asterisk consist of a low threshold differ-ential discriminator and a high level leading edge discriminator. The depictedtrigger logic shows the effective logic obtained by not using all input channelsof the trigger 1A module.

3.4. DETECTORS 63

Due to finite acceptances and resolutions there is for all true coincidences(that is, for events where the hits in the various detectors indeed correspondto particles emerging from one and the same scattering event) a finite spreadin the times of arrival of the ATRs. The time window for coincidences shouldbe wider than this spread, in order to estimate the amount of random coinci-dences, resulting in time windows of the order of 100 ns. This puts a constrainton the maximum acceptable individual ATR rates in the CD.

Therefore, the ATR rate of the Neutron Detector was reduced by gating itsfirst level trigger with the pretrigger signal of the BigBite spectrometer. Thisreduced the ND ATR rate from the order of 100 kHz to below 50 Hz. The Big-Bite pretrigger signal is defined by the fast signals of its scintillator and theCerenkov detector. For real coincidences the timing was such that the fastestneutrons arrived 50 ns after the start of the pretrigger gate. The BigBite pre-trigger is not equivalent to a BigBite ATR: only 30–40% of the gated NeutronDetector ATRs resulted in an ETR for a BigBite - Neutron Detector coincidence.


Documents

Experimental setup - Universiteit Utrecht